\begin{table}

\caption{\label{tab:dictionary_comparison}Correlations of CHES anti-elite salience estimates with measurements created with our XLM-T classifier, the dictionary compiled by \citet{grundl_populist_2020}, and the dictionary compiled by \citet{rooduijn_measuring_2011}, respectively. Estimates computed on German-language tweets in Austrian, German, and Swiss parties tweets published in the 12 months prior to the respective CHES waves' field end date. }
\centering
\fontsize{9}{11}\selectfont
\begin{tabular}[t]{cccc}
\toprule
\multicolumn{2}{c}{ } & \multicolumn{2}{c}{Dictionary} \\
\cmidrule(l{3pt}r{3pt}){3-4}
CHES wave & XML-T classifier (ours) & Gründl (2020) & Rooduijn and Pauwels (2011)\\
\midrule
2014 & 0.491 & 0.199 & 0.341\\
2019 & 0.665 & 0.232 & 0.569\\
\bottomrule
\end{tabular}
\end{table}
